flink如何根据时间过滤数据？

Question

我有一些类似这种格式的数据：

(id, time, value)

给出以下仿真数据（可能有重复数据）：

("a1", "2022-06-28 00:00:00", "0.23"), // The time interval is 15 minutes, and there is only 24-hour data of the day
("a1", "2022-06-28 00:15:00", "0.89"),
...
("a1", "2022-06-28 23:59:59", "0.11"),

("b1", "2022-06-28 00:00:00", "0.23"), 
("b1", "2022-06-28 00:15:00", "0.89"),
...
("b1", "2022-06-28 23:59:59", "0.11"),


("c1", "2022-06-28 00:00:00", "0.23"), 
("c1", "2022-06-28 00:15:00", "0.89"),
...
("c1", "2022-06-28 23:59:59", "0.11"),

假设现在是2022-06-28 16:00:00 ，我想计算 1h,45min,30min,15min 之前和现在的数据。

输出应该是这样的：

("a1", "2022-06-28 15:00:00", "1"),
("a1", "2022-06-28 15:15:00", "1"),
("a1", "2022-06-28 15:30:00", "1"),
("a1", "2022-06-28 15:45:00", "1"),
("a1", "2022-06-28 16:00:00", "1"),
("b1", "2022-06-28 15:00:00", "1"),
("b1", "2022-06-28 15:15:00", "1"),
("b1", "2022-06-28 15:30:00", "1"),
("b1", "2022-06-28 15:45:00", "1"),
("b1", "2022-06-28 16:00:00", "1"),
("c1", "2022-06-28 15:00:00", "1"),
("c1", "2022-06-28 15:15:00", "1"),
("c1", "2022-06-28 15:30:00", "1"),
("c1", "2022-06-28 15:45:00", "1"),
("c1", "2022-06-28 16:00:00", "1"),

如何编写 Flink 程序？ 最好用 Java 或 Scala 编写。 如果您能给我看一些代码片段，我将不胜感激！

Answer 1

我建议看一下 Flink 的 Windowing Table Valued Functions。 您可以在https://nightlies.apache.org/flink/flink-docs-master/docs/dev/table/sql/queries/window-tvf/#tumble找到文档和示例